Policies such as the SEC’s Fair Disclosure Rule, and technologies such as SEC EDGAR, aim to disseminate corporate disclosures to a wider audience. In this study, we adopt an experimental approach to measure whether this wider disclosure is beneficial to investors. Analytical predictions based on theories of non-revealing and full-revealing prices can be summarized as “None > All > Half”. A laboratory study was conducted to test these predictions. Subjects’ preference for the fraction of informed traders can be summarized as “Half > None > All”, i.e., investors most favor a situation where a random half of investors are informed. We explore two possible explanations for the contradiction of the theoretical predictions, either that the analytical models fail to predict market behavior, or that they succeed in predicting market behavior but nevertheless fail to predict subjects’ preferences for the different sets of risk faced in each market. We ultimately adopt the second approach, and propose that subjects have different attitudes toward different sources of risk, a phenomenon which traditional analytical models do not consider.